In packet communication, a data size transmissible at a time is decided beforehand. Therefore, when a packet of a larger size than the transmissible size is transmitted, the packet has to be divided into a plurality of fragments.
A variety of ways of fragment size decision have been proposed. However, a fragment size in consideration of upper layer characteristics has not been studied.